ANALISIS SENTIMEN OPINI PENGGUNA TWITTER PADA APLIKASI BIBIT MENGGUNAKAN MULTINOMIAL NAÏVE BAYES

The COVID-19 pandemic has increased the interest number of capital market investors in Indonesia. One of the factor in Indonesian’s investment interest is the emergence of fintech in the investment sector. One of the fintech companies in mutual fund investment is "Bibit". To find out user opinions on the Bibit application, a sentiment analysis was carried out on Twitter’s users. This study aims to analyze the sentiments of twitter users' opinions on the Bibit application using a combination of Lexicon-Based and Multinomial Naïve Bayes methods. The training data used were 2211 tweets and the validation data was 553 tweets. In the model training process, the training accuracy level is 91.50% and the validation accuracy rate is 85.35%. Model testing was carried out using 39 new tweet data and obtained an accuracy rate of 88%. Sentiment analysis using this method is visualized in the form of pie charts, graphs, and wordclouds. Based on the results of visualization of Twitter social media user sentiment towards the seedling application, it tends to be positive with a percentage of 52% positive and 48% negative.